Sign in
← AI PulseJul 27, 2026

Deep · news · Multi-source brief

Vercel AI SDK Adds Experimental Speech Translation and TogetherAI Usage Reporting

Vercel's AI SDK introduces an experimental speech translation model specification and streaming speech-to-speech translation, alongside enabling token usage reporting for TogetherAI streaming responses.

By Illumora Editorial

Source · Jul 27, 2026, 9:55 PM · On Illumora · Jul 27, 2026, 10:08 PM

Media from the primary source — shown here so you can stay on Illumora.

Synthesized from multiple allowlisted primaries on the same event. Lanes →

Brief drafted by Illumora’s editorial model from the linked primary source. Ops desk reviews flagged pieces. How we write →

Read the source →Vercel AI SDK Changelog — Release @ai-sdk/xai@4.0.19 · vercel/ai
Save

The Vercel AI SDK has released updates, including an experimental speech translation model specification and a feature for streaming speech-to-speech translation. These additions are part of the ai@7.0.38 patch changes, which also address client tool errors.

Separately, the @ai-sdk/togetherai@1.0.50 package now includes functionality to report token usage for streaming responses from TogetherAI. This update aims to provide more transparency regarding resource consumption during AI model interactions.

Key Points

  • The Vercel AI SDK ai@7.0.38 patch changes include an experimental speech translation model specification, Experimental_SpeechTranslationModelV4.
  • The SDK now supports experimental_streamTranslate for streaming speech-to-speech translation.
  • Client tool errors for invalid provider-executed tool calls are now handled to avoid synthesizing them.
  • The @ai-sdk/togetherai@1.0.50 package enables includeUsage for TogetherAI, allowing streaming responses to report token usage.
  • Updated dependencies across various SDK packages, including @ai-sdk/gateway@4.0.29, @ai-sdk/provider@4.0.4, and @ai-sdk/provider-utils@5.0.13.

Context

According to the Vercel AI SDK changelog, the updates to ai@7.0.38 introduce new capabilities for speech processing within the SDK. This includes both a new model specification and a streaming translation feature. The 7bd6bdd patch specifically addresses an issue where client tool errors were being synthesized for invalid provider-executed tool calls, improving error handling.

Separately, the @ai-sdk/togetherai@1.0.50 release, noted on 27 Jul, focuses on enhancing observability for TogetherAI users. The 5dcbf7e patch enables the includeUsage option, which allows developers to receive token usage reports during streaming interactions with TogetherAI models.

Why It Matters

These updates offer builders new experimental capabilities for integrating speech translation into their applications and provide more detailed usage metrics for TogetherAI models. The inclusion of token usage reporting for streaming responses can assist developers in monitoring and optimizing their AI application costs and performance.

What To Do

  • Review the Vercel AI SDK changelog for ai@7.0.38 to understand the full scope of speech translation features.
  • Test the Experimental_SpeechTranslationModelV4 and experimental_streamTranslate functions in development environments.
  • For TogetherAI users, verify that includeUsage is enabled in @ai-sdk/togetherai@1.0.50 to monitor token consumption.
  • Note the updated dependency versions across various @ai-sdk packages and plan for necessary upgrades.